| 基于注意力机制和机器视觉图像的地铁安检违禁物检测方法 |
 点此下载全文 |
| 引用本文：熊振兴1，姜旭2.基于注意力机制和机器视觉图像的地铁安检违禁物检测方法[J].计算技术与自动化,2025,(4):116-121 |
| 摘要点击次数: 8 |
| 全文下载次数: 2 |
|
|
| 中文摘要:为了保证地铁乘客安全以及地铁的顺利运行，提出了基于注意力机制和机器视觉图像的地铁安检违禁物检测方法。采用机器视觉图像处理技术中的中值滤波技术，去除X射线安检设备获取的地铁安检物品图像中的椒盐噪声后，通过Canny边缘检测算子获取去除椒盐噪声的地铁安检物品图像的边缘信息。将地铁安检物品边缘图像输入到YOLOv5s检测模型中，通过引入注意力机制和特征融合模块实现违禁物的特征信息增强和提取违禁物融合特征的过程，通过对检测模型的训练，最终实现地铁安检违禁物检测。通过实验验证，该方法能够获得较为精确的地铁安检违禁物检测结果，对小目标和液体的违禁物检测精度较高，能够通过红色标识框实现违禁物的标识，对地铁安检过程提供了便利。 |
| 中文关键词:注意力机制 机器视觉 地铁安检 违禁物检测 YOLOv5s算法 特征融合 |
| |
| Detection Method of Contraband in Subway Security Check Based on Attention Mechanism and Machine Vision Image |
|
|
| Abstract:In order to ensure the safety of subway passengers and the smooth operation of the subway, a method for detecting prohibited items in subway security checks based on attention mechanism and machine vision images is proposed. The median filtering technology in machine vision image processing is used to remove the salt and pepper noise in the subway security check item images obtained by X-ray security check equipment. The edge information of the subway security check item images with salt and pepper noise removed is obtained through the Canny edge detection operator. Input the edge images of subway security check items into the YOLOv5s detection model, enhance the feature information of prohibited items and extract the fusion features of prohibited items by introducing attention mechanism and feature fusion module. Through training the detection model, the detection of subway security check prohibited items is ultimately achieved. Through experimental verification, this method can obtain more accurate detection results of prohibited items in subway security checks, with high accuracy in detecting small targets and liquids. It can achieve the identification of prohibited items through red boxes, providing convenience for the subway security check process. |
| keywords:attention mechanism machine vision subway security check detection of prohibited substances YOLOv5s algorithm feature fusion |
| 查看全文 查看/发表评论 下载pdf阅读器 |
次